Modeling Thermodynamic Properties of Isomeric Alkanes with a New

Subscriber access provided by READING UNIV

Article

Modeling thermodynamic properties of isomeric alkanes with a new branched equation of state Yuchong Zhang, and Walter G Chapman Ind. Eng. Chem. Res., Just Accepted Manuscript • DOI: 10.1021/acs.iecr.7b03951 • Publication Date (Web): 04 Jan 2018 Downloaded from http://pubs.acs.org on January 4, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Industrial & Engineering Chemistry Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Industrial & Engineering Chemistry Research

Modeling thermodynamic properties of isomeric alkanes with a new branched equation of state Yuchong Zhang and Walter G. Chapman∗ Department of Chemical and Biomolecular Engineering, Rice University, Houston E-mail: [email protected]

Abstract In this paper we present an extension of the statistical associating fluid theory (SAFT) for branched molecules with a Lennard Jones dimer reference fluid (SAFTD-LJ-Branch). The theory successfully predicts how branched architecture affects the attraction and repulsion between molecules. SAFTD-LJ-Branch takes a form similar to SAFTD-LJ with an additional parameter N B introduced to account for the branching effect. We propose an approach relating N B to the number of different types of articulation segments. The theory is used to study the effect of chain architecture on the thermodynamic properties of isomeric alkanes. SAFTD-LJ-Branch accurately predicts the phase diagram of pure butane, pentane or hexane isomers. Further, vapor pressures of ntriacontane and squalane are predicted without further fitting and shown to be in semiquantitative agreement with experimental data. Finally, SAFTD-LJ-Branch is demonstrated to be well applicable to mixtures as we model the vapor-liquid coexistence of binary alkane mixtures containing different hexane isomers and recover the experimental trends.

ACS Paragon Plus Environment

1

Industrial & Engineering Chemistry Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

Introduction

Page 2 of 17

for off-lattice applications. Another successful category of models is perturbation theory, particularly, the Statistical Associating Fluid Theory (SAFT) equation of state (EoS). 21–25 SAFT models molecules as chains of spherical segments. It lays its foundation on Wertheim’s first order thermodynamic perturbation theory (TPT1), 26–29 which is a rigorous statistical mechanical theory of associating spheres. The excess Helmholtz free energy of a system can be viewed as a sum of the reference fluid contribution plus contributions from chain formation, association, dispersion and other interactions. The perturbation expansion is typically written for a hard sphere reference fluid, 21 Lennard-Jones reference fluid 30 or square well fluid. 31 Depending on the applied interaction potential, different versions of SAFT have been developed, such as PC-SAFT, 32,33 Soft-SAFT 34 and SAFT-VR. 35 In TPT1, only pair correlations are included in the cluster expansion for association or chain formation. Therefore, interactions between non-nearest neighbors are neglected. As a result, all the major versions of SAFT EoS are implicitly developed for linear chains. To explicitly include the structural information in the SAFT EoS, a higher level of perturbation theory must be considered, at least a second order perturbation (TPT2). 36 TPT2 was extended from TPT1 by Wertheim, where all graphs with a single path of two attraction bonds are retained in addition to the graphs retained in TPT1, thus providing information between bonded pairs and next nearest neighbors along the chain. TPT2 differs from TPT1 by a small term and achieves improved agreement with simulation results. Phan et al. 37 firstly derived a TPT2 equation of state for hard sphere chains, including star-like molecules. They were able to predict the pressure of freely rotating chains and that of trimers as a function of the bond angle. Marshall and Chapman 38 extended this approach further to generate a TPT2 correction for any branched molecule. However, it is challenging to evaluate the TPT2 contributions in the above equations of state. Besides referring to TPT2, another strategy to improve the behavior of the SAFT EoS is

Molecular architecture affects the phase behavior of polyatomic molecules. 1 Energetically, the intermolecular interactions between atoms or groups can be shielded due to the addition of branches. From the entropic respect, different favored conformations might be assumed for different molecular structures. 2 All these factors lead to a difference in phase change, fluid structure as well as other thermodynamic properties in a complex way. 3–6 On the other hand, whether it is short chains or long polymers, any property change due to branching matters for industrial processing. The effects of molecular architecture on thermodynamic properties are obtained primarily from experiments, which are expensive and time consuming. It is impossible to consider the full range of state points and branched structures only from experiments. Practicing engineers many times must estimate fluid properties for components lacking experimental data. This is sometimes done by comparing with reference components that differ in molecular architecture. Therefore, a theoretical model that accurately predicts the effect of branching on thermodynamic properties is needed. One popular theory to describe, how molecular structure affects fluid properties, is lattice cluster theory (LCT), originally developed by Freed et al. 7,8 and extended to compressible systems by Dudowicz et al. 9,10 . LCT uses a double series expansion in the inverse coordination number and the reduced interaction energy to arrive at the free energy of a structured lattice fluid characterized by several combinatorial numbers giving the number of distinct ways to find a given substructure in the molecular architecture of the fluid under consideration. It has been tested against lattice Monte Carlo simulation 11 and extended to include semi-flexibility and association interactions in the same lattice formalism. 12 Especially the group of Enders has applied it successfully for polymeric and non-polymeric systems. 13–20 However, as a lattice theory, LCT sometimes performs weakly for the gas phase. Also the selection of the lattice coordination number can be quite empirical


2

Page 3 of 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


2

using a fluid of dimers as reference fluid. The dimers consist of two spheres bonded at contact and it is assumed that the correlation function for longer chains can be approximated by the dimer correlation function. As a result, in this dimer scheme, a segment knows the information of nearest neighbor and also the next nearest neighbor. The first SAFT-D was independently developed by Ghonasgi and Chapman 39–42 and by Chang and Sandler 43 with the reference fluid being hard sphere dimers. With the same strategy, the Lennard Jones (LJ) reference fluid can also be extended to LJ dimer and Johnson 44 used it to predict thermodynamics of pure components. Later, Blas and Vega 45 also demonstrated that Soft-SAFT-D could be applied to mixtures with improvement over Soft-SAFT, which basically uses LJ spheres as the reference fluid. Despite the improvement over the monomer version, no branching effect is taken into consideration for the dimer version. Recently, Marshall and Chapman 46 proposed a different approach to construct the second order correction to the free energy for branched molecules and showed that a similar form of correction can be applied in a TPT1-D type of equation of state. Specifically, a correction to SAFTD-LJ will yield a new equation of state that not only maintains a simple analytic form but also directly applies to real components. In this work, we follow similar lines and find it possible to further modify the correction to improve predictions for real systems. We name this equation of state as SAFTDLJ-Branch. We have used SAFTD-LJ-Branch to systematically study the phase behavior of alkane isomers. The phase diagram and vapor pressures of pure isomeric alkanes as well as vapor liquid equilibrium (VLE) of alkane mixtures are modeled. The theoretical predictions agree well with experimental data, validating the new equation of state. The paper is organized as follows. In Sec. 2, SAFTD-LJ-Branch is developed for branched molecules. In Sec. 3, parameters are fitted, correlated and validated. Alkane systems of interest are studied with the application of SAFTDLJ-Branch.

Theory

In this section, we introduce SAFTD-LJ, 44 and then include the branch correction to obtain SAFTD-LJ-Branch. Finally, the conventional van der Waals one-fluid mixing rules (VdW1) 47 are applied for extension to mixtures.

2.1

Formalism of SAFTD-LJ

In the scheme of SAFTD-LJ, if no association exists, the residual Helmholtz free energy of a system can be written as the sum of LJ sphere free energy and chain formation free energy. Ares = As + Achain

(1)

Three parameters are needed in this model, that is, the number of spherical LJ segments in a molecule m, dispersion energy between LJ segments and segment diameter σ. and σ are explicitly included in the LJ potential. h σ σ i (2) uLJ (r) = 4 ( )12 − ( )6 r r The term As accounts for the contribution from Lennard Jones segments and the term Achain accounts for the contribution from the effect of segments connected into dimers and then chains of fixed number. We also define ρ as number density, N as number of chain molecules or segments, V as volume, T as temperature, p as pressure and µ as chemical potential. The subscripts s and c refer to segment and chain respectively. In the following formulas, all the quantities will be expressed in reduced forms σ 3 , T ∗ = kT , p∗ = as below: ρ∗ = ρσ 3 = N V pσ 3 µ , A∗ = NAkT , µ∗ = kT . Now we have the reduced residual Helmholtz free energy. A∗res = mA∗s + A∗chain

(3)

Note that the existence of m is due to the fact that A∗s is reduced with the number of segments while A∗res and A∗chain are reduced with the number of chains. A∗s is proposed by Johnson et al. 48 through correlations of simulation results, and its expression can be found in the appendix.


3


(a)

a segment not only knows its nearest neighbor, but also the next nearest neighbor. Thus this term should be proportional to the number of three consecutive segments that can be found in the chain. For a linear chain with m segments, this number is m − 2. For branched chains, we come up with a way to determine this number through observation of articulation segments, that is, the segments with three or more arms. Fig. 1 (a) is a branched chain with 6 segments and there are two 3-arm articulation segments, colored as grey. The number of three consecutive segments is 6. On the other hand, Fig. 1 (b) is also a branched chain with 6 segments and there is one 4-arm articulation segment, colored as black. The number of three consecutive segments is 7. Generalizing to a branched chain with m segments, we introduce a new parameter N B to reflect the difference in the number of three consecutive segments compared with that of linear isomer (which is m − 2). By rule a 3arm articulation segment contributes 1 to N B and a 4-arm articulation segment contributes 3 to N B. For molecules with multiple articulation segments, N B is additive. Actually, this methodology can also be applied to articulation segments with more arms. However, for the molecules we are most interested in, such as alkanes, the inclusion of 3-arm and 4-arm articulation segments are enough. Based on this analysis, we develop a new chain term which takes branched structure into consideration.

(b)

Figure 1: Molecules consisting of six segments but with different structures. Molecule (a) has two 3-arm articulation segments (grey) and molecule (b) has one 4-arm articulation segment (black). Each grey segment receives an N B = 1 and the black segment receives an N B = 3.

A∗chain is expressed as 44,45 A∗chain =

Achain m = − ln[gs (σ)] Nc kT 2 m + (1 − )ln[gd (σ)] (4) 2

Where gs and gd are the pair radial distribution function of LJ spheres and the site-site distribution function of LJ dimers respectively. Expressions for gs and gd at contact are also obtained from correlations fit to molecular simulation; 44,49 see details in the appendix.

2.2

Formalism Branch

of

SAFTD-LJ-

Since the term As has no dependence on how the LJ spheres are connected and arranged, to account for branching, we focus on the chain formation term Achain . The reduced chain term in SAFTD-LJ can also be written as A∗chain =

Page 4 of 17

A∗chain =

Achain = (1 − m)ln[gs (σ)] Nc kT m−2 gd (σ) −( )ln (5) 2 gs (σ)

Achain = (1 − m)ln[gs (σ)] Nc kT m − 2 + NB gd (σ) −( )ln (6) 2 gs (σ)

When N B = 0, this term is the same as Eq. 5, meaning that SAFTD-LJ-Branch simply reduces to SAFTD-LJ for linear chains. This expression is the same as proposed by Marshall and Chapman. 46 They applied the theory to branched alkanes, but not to mixtures. It is interesting to note that the architecture in LCT is accounted for by a series of combinatorial numbers, 12,18 the number of segments m, the number of bonds m − 1, the number of two

The first term in the bracket is the chain contribution in SAFT-LJ, 30,49–51 and the second term is a correction term to incorporate structural information for dimers. If there is no difference between the LJ sphere correlation function and LJ dimer correlation function, then this term vanishes. In SAFTD-LJ,


4

Page 5 of 17

consecutive bonds N2 and the number of three consecutive bonds N3 . The impact of architecture on thermodynamics arises starting with the occurrence of N2 in the series expansion of LCT, which is on the same level of approximation as our expression m−2+N B. This similarity between the two theories is especially striking, because they are derived in a completely different formalism. This strengthens both theoretical results to some extent and it would be interesting to see, if further similarities occur, once trimer and tetramer correlation functions are available. The resulting SAFTD-LJ-Branch, has four parameters in total, that is m, σ, and N B. We assume that to model isomers with different branching structure, m, σ and can be kept the same and the only different parameter is N B. This assumption allows us to predict the phase behavior of numerous chemicals from data for isomers. The approach should be applicable to isomers with similar numbers and types of functional groups, such as alkanes and primary alcohols.

5 4.5 4

m

3.5 3 2.5 2 1.5 1 20

40

60

80

100

120

140

160

100

120

140

160

100

120

140

160

MW

(a) 300

250

mσ3

200

150

100

50 20

40

60

80

MW

2.3

(b)

VdW1 theory for mixtures 280

Like other SAFT equations of states, SAFTDLJ-Branch can be easily extended to mixtures with the approximation of van der Waals OneFluid Theory, 47 which is a well-established conformal solution theory. VdW1 defines parameters of a hypothetical pure fluid x having the same residual properties as the mixture of interest. For σx and x , the mixing rules are: P P 3 i j mi mj xi xj σij 3 P P (7) σx = i j mi mj xi xj P P 3 i j mi mj xi xj ij σij 3 x σx = P P i j mi mj xi xj

270 260 250

ǫ/k

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


240 230 220 210 200 20

40

60

80

MW

(c)

(8)

Figure 2: Parameters of SAFTD-LJ EoS for the series of n-alkanes, correlated from propane to n-decane. Lines are for eye guidance. (a) m as a function of M W . (b) mσ 3 as a function of M W . (c) /k as a function of MW.

σii + σjj √ , ij = ii jj (9) 2 Where xi is the mole fraction of component i. The effective segment number or branching parameter of the conformal fluid are simply the average segment number or branching parameσij =

ter of the mixture. X mx = mi xi , i


5

N Bx =

X i

N B i xi

(10)


Page 6 of 17

Table 1: Pure component parameters for normal alkanes from C3 to C10 carbon number 3 4 5 6 7 8 9 10

3

m 1.913 2.249 2.581 2.941 3.276 3.651 3.993 4.335

σ[˚ A] 3.726 3.819 3.874 3.904 3.926 3.933 3.953 3.971

/k[K] 214.3 232.1 245.3 253.9 261.8 266.7 271.5 275.3

AAD(pV )% 0.535 1.256 0.877 1.057 0.628 0.859 0.990 0.868

Results and Discussion

T range[K] 200-350 195-420 220-460 250-500 229-501 254-526 266-554 286-574

pressions.

Despite their simplicity compared to other molecules, the alkane families are of importance across many industries. The thermodynamic properties of even light alkane isomers are difficult to predict from existing models. Therefore we focus on the phase behavior of alkane isomers to test the accuracy of SAFTDLJ-Branch. The following results begin with fitting and correlation of parameters, and then SAFTD-LJ-Branch is applied to pure components as well as alkane mixtures.

3.1

AAD(vL )% 0.495 2.076 1.566 1.817 0.814 0.611 0.583 0.570

m = 0.0248M W + 0.8063 mσ 3 = 1.7437M W + 23.469 314.998M W − 209.288 /k = M W + 19.7705

(11a) (11b) (11c)

3.2

Phase equilibrium of pure isomeric alkane

3.2.1

VLE of n-alkanes versus i-alkanes

In the frame of SAFTD-LJ-Branch, isomeric alkanes share the same parameters except for N B, which is related to the structure of a molecule and determined by the numbers of different types of articulation points. As a first examination, we model vapor liquid equilibrium for n-alkanes and their corresponding i-alkane isomers. In Fig. 3, the predictions for butane, pentane and hexane are compared against NIST values, 52 which are correlations of experimental data. We observe overall good agreement. The model does extremely well in predicting vapor density and vapor pressure while it underestimates the branching effect on liquid density. Table. 2 shows the average absolute deviations for predictions of vapor pressure and liquid molar volume in the investigated temperature range. In agreement with experiment, we find that i-alkanes tend to have a higher critical density and a lower critical temperature than its linear isomer. The error in critical temperature can be corrected by including fluctuation. 54 Significant deviations are expected if the molecular structure becomes spherical, such as i-butane. This is because the

Parameters for Pure normal alkanes

As mentioned before, SAFTD-LJ-Branch simply reduces to SAFTD-LJ for linear chains with N B = 0 and we should have the same parameters (m, σ and ) for the two equations of state. However, previous works with SAFTD-LJ 44,45 mainly consider a LJ model fluid and no parameters are available for real components. Therefore, we started by fitting parameters for normal alkanes from C3 to C10 . During the fitting process, m, σ and are fitted at the same time to vapor pressure (pV ) and liquid molar volume (vL ) data of pure components 52 within the temperature range of interest. The fitting results are presented in Table. 1. The equation of state parameters are plotted versus molecular weight (M W ) of n-alkanes starting from propane in Fig. 2. We see that the plots of all three parameters show a smooth trend. Inspired by Pedrosa et al., 53 we succeed in correlating the parameters with molecular weight with the following ex-


6

Page 7 of 17

4

450

400

3 350

T[K]

Log(P LV[KPa])

3.5

2.5

300 2 250 1.5

200

1 2

2.5

3

3.5

4

0

4.5

2000

4000

6000

8000

10000

12000

ρ[mol/m3 ]

1000/T[K]

(a)

(b)

4

500

3.5

450

400 2.5

T[K]

Log(P LV[KPa])

3

350

2 300 1.5 250

1

200

0.5 2

2.5

3

3.5

4

0

4.5

2000

4000

6000

8000

10000

ρ[mol/m3 ]

1000/T[K]

(c)

(d)

4

550

3.5

500

3

450

2.5 400

T[K]

Log(P LV[KPa])

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


2

350 1.5 300 1 250

0.5

200

0 2

2.5

3

3.5

4

0

4.5

2000

4000

6000

8000

ρ[mol/m3 ]

1000/T[K]

(e)

(f)

Figure 3: Vapor pressures (a, c, e) and temperature/density coexistence curves (b, d, f) for n-butane and i-butane (a, b), n-pentane and i-pentane (c, d), n-hexane and i-hexane (e, f). Squares give NIST values of n-alkanes and triangles give NIST values of i-alkanes. 52 Solid lines give theoretical results of n-alkanes and dashed lines give theoretical results of i-alkanes.


7


Page 8 of 17

Table 2: Performance of SAFTD-LJ-Branch for pure components

i-butane i-pentane 2-methylpentane 3-methylpentane (2,3)-dimethylbutane (2,2)-dimethylbutane n-triacontane squalane

AAD(pV )% 8.00 3.55 3.56 5.58 13.20 10.33 14.89 32.51

AAD(vL )% 3.00 0.52 0.45 1.09 1.35 1.99 2.54 4.29

branching effect is accounted on the basis of a linear chain model, and increasing sphericity introduces unanticipated symmetry, which affects the thermodynamics of molecules. 3.2.2

T range[K] 220-387 230-438 249-473 249-359(pV ), 343-473(vL ) 244-354(pV ), 343-473(vL ) 237-346(pV ), 343-463(vL ) 450-780 450-780

effect on the vapor density while it underestimates that on the liquid density. Moreover, 2methylpentane and 3-methylpentane show nonnegligible differences from the correlations, especially for the liquid density. This indicates that more structural information needs to be included to accurately predict densities of isomers with similar extent of branching. For the prediction of vapor pressures, the theory seems to be adequate. The tendency of critical properties for different isomers are found to be consistent with experimental data except for the critical temperature of 2-methylpentane, which is exceptionally low experimentally compared with (2,3)-dimethylbutane. The quantitative summary of performance of SAFT-LJ-Branch for different branched hexane isomers is presented in Table. 2. Some may doubt that N B can be determined simply from the structure of hexane since the molecule is modeled as made up of 2.941 effective segments (instead of the carbon number in the backbone) and N B should also be an effective branching parameter. For the modeling of (2,2)-dimethylbutane, we notice that in Fig. 5 the vapor side displays unexpected curvature approaching the critical temperature. We believe that this behavior is a result of the relatively large value of N B(= 3) compared with m. Further, we note that a slightly higher or smaller value of N B can improve the agreement with vapor pressure depending on the component. However, since N B assigned this simple way generates encouraging results, we stick to the parameter values in Table. 3 for most of this work. We will discuss the role of N B later where it is not necessarily an integer.

VLE of hexane isomers

To validate the potential of SAFTD-LJ-Branch being applied to different structures that an alkane isomer may assume, hexane is considered a good choice since it has five isomers. Among the five hexane isomers: n-hexane is linear; 2methylpentane and 3-methylpentane both have one branch but at different carbon positions; (2,3)-dimethylbutane has two branches located at different positions on the backbone while the two branches of (2,2)-dimethylbutane connect to the same carbon atom. Following the argument in Sec. 2.2, it is straightforward to determine the values of N B for each hexane isomer, as is listed in Table. 3; the values of parameters m, σ and are the same as for n-hexane. As shown in Fig. 4, vapor pressures are modeled for each hexane isomer and plotted as a function of reciprocal temperature. Theoretical results match the NIST values 52 well, especially at low temperatures. Note that using SAFTD-LJ-Branch, 2-methylpentane and 3-methylpentane share the same set of parameters and should yield the same results theoretically. While not identical, the experimental vapor pressures of these two isomers fall very close to each other. Saturated densities at different temperatures for each hexane isomer are plotted in Fig. 5 and compared against the values of Kay 55 correlated from experiments. The agreement for densities is not as satisfactory as that for vapor pressures. Overall, the theory overestimates the branching


8

Page 9 of 17

Table 3: Parameters of different isomers of hexane σ[˚ A] 3.904 3.904 3.904 3.904 3.904

m 2.941 2.941 2.941 2.941 2.941

n-hexane 2-methylpentane 3-methylpentane (2,3)-dimethylbutane (2,2)-dimethylbutane

/k[K] 253.9 253.9 253.9 253.9 253.9

bone. To model n-triacontane, we extrapolate the parameters from Eq. 11 and we obtain m = 11.292, σ = 4.069˚ A, /k = 300.454K. We use those same numbers for the parameters of squalane with an additional N B = 6. Fig. 6 shows the theoretical predictions for vapor pressures of both n-triacontane and squalane, compared aginst values from NIST. 52 We find the theory does a good job in matching the data at high temperatures, particularly since the parameters are extrapolated. At low temperatures, the predicted vapor pressures of both chemicals begin to deviate from the data points with squalane showing a higher deviation. The average absolute deviations for predicted vapor pressures of n-triacontane and squalane are 14.89% (maximum is 27.67%) and 32.51% (maximum is 61.71%). These numbers are large partly due to the small values of the true vapor pressures when temperature is low. However, the same direction of deviations observed for both chemicals implies a successful capture of the branching effect. Due to the unavailability of vapor densities from literature, only the saturated liquid densities of the two isomers are plotted in Fig. 7. Contrary to vapor pressures, the theoretical results are more accurate at low temperatures and begin to deviate from NIST values approaching the critical point. There is little difference between the liquid densities of n-triacontane and squalane at low temperatures and as a matter of fact, the theory incorrectly predicts a slightly higher density for squalane until a crossover at high temperature. The inconsistency may be a result of the parameters extrapolated from short chain alkanes or deficiencies of SAFTDLJ-Branch itself. The performance of SAFTDLJ-Branch for n-triacontane and squalane is summarized in Table. 2. Despite defects, the results are still inspiring since the parameters are extrapolated from decane and the only required input is the structure of the chemical. This new equation of state provides semi-quantitatively accurate predictions of the effect of molecular structure for heavier molecules.

NB 0 1 1 2 3

3 n-hexane 2-methylpentane 3-methylpentane 2,3-dimethylbutane 2,2-dimethylbutane

Log(P LV[KPa])

2.5

2

1.5

1

0.5

0 2.5

3

3.5

4

4.5

1000/T[K]

Figure 4: Log base 10 vapor pressures for different isomers of hexane versus reciprocal temperature. Symbols give NIST values 52 and lines are theoretical predictions. Upper lines are of higher N B. 600 n-hexane 2-methylpentane 3-methylpentane 2,3-dimethylbutane 2,2-dimethylbutane

550

500

T[K]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


450

400

350

300 0

1000

2000

3000

4000

5000

6000

7000

8000

ρ[mol/m3 ]

Figure 5: Temperature/density coexistence curves for hexane isomers. Symbols give correlation values of Kay 55 and lines are theoretical predictions. Lower lines are of higher N B.

3.2.3

VLE of squalane

n-triacontane

versus

In this section we test our theory for heavier alkanes. One chemical of interest is squalane, an isomer of n-triacontane. It has six methyl branches almost evenly spread over its back-


9


4.5

4 n-triacontane squalane

3

4 3.5

2 1

P[MPa]

Log(P LV[KPa])

3

0

2.5 2 1.5

-1

1

-2 0.5

-3 0 0

-4 1

1.2

1.4

1.6

1.8

2

2.2

0.2

0.4

0.6

0.8

1

0.8

1

0.8

1

xpropane,y propane

2.4

1000/T[K]

(a)

Figure 6: Log base 10 vapor pressures of n-triacontane and squalane versus reciprocal temperature. Symbols give NIST values. 52 Solid line gives theoretical results of n-triacontane and dashed line gives theoretical results of squalane.

4.5 4 3.5

P[MPa]

3

1800 n-triacontane squalane

1600

2.5 2 1.5 1 0.5

1400

ρ[mol/m3 ]

0 0

1200

0.2

0.4

0.6

xpropane,y propane

(b)

1000 4.5

800

4

600 400

500

600

700

800

3.5

900

T[K]

3

P[MPa]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 17

Figure 7: Saturated liquid densities of n-triacontane and squalane versus temperature. Symbols give NIST values. 52 Solid line gives theoretical results of ntriacontane and dashed line gives theoretical results of squalane.

2.5 2 1.5 1 0.5 0

3.3 3.3.1

0

Phase equilibrium of alkane mixtures VLE of propane

isomeric

hexane

0.2

0.4

0.6

xpropane,y propane

(c)

and

Figure 8: VLE of the mixture propane + isomeric hexane at 348.15K (lower curver) and 373.15K (upper curves). In each figure, solid lines are theoretical reWe further extend the pure component syssults for mixture propane + n-hexane and dashed lines tem to mixtures. Vapor liquid equilibrium are theoretical results for mixture propane + branched hexane. Squares give experimental results for mixof isomeric hexane (using parameters in Tature propane + n-hexane and triangles give experible. 3) and propane are studied at two temmental results for mixture propane + branched hexperatures (348.15K and 373.15K), as shown in ane. 4 (a) Comparison with mixture propane + (2/3)Fig. 8. For a better look, the system of nmethylpentane. (b) Comparison with mixture propane hexane and propane is compared against the + (2,3)-dimethylpentane. (c) Comparison with mixture propane + (2,2)-dimethylpentane. ACS Paragon Plus Environment

10

Page 11 of 17

system of (2/3)-methylpentane and propane, system of (2,3)-dimethylbutane and propane, system of (2,2)-dimethylbutane and propane, respectively. In general the model gives a good description of the phase equilibrium compared with experimental data 4 and the branching effect is particularly clear for the dew curve. At 373.15K, the average deviation for the predictions of system of (2,2)-dimethylbutane and propane is 1.90% (maximum is 6.13%) in bubble pressure and 12.06% (maximum is 22.34%) in dew pressure. The systems of propane and other hexane isomers behave similarly or better. At low mole fraction of propane, the theory tends to overestimate pressures while at high mole fraction of propane underestimation occurs. However, the overall tendency is nicely captured. With the same composition, the system with more branching (higher N B) features higher dew pressure and bubble pressure, consistent with the pure component case. 3.3.2

60

50

P[KPa]

40

30

20

10

0 0

0.2

0.4

0.6

0.8

1

x2-methylpentane,y 2-methylpentane

Figure 9: VLE of the mixture 2-methylpentane + noctane at various tmeperatures. Solid lines are theoretical predictions with N B = 1 for 2-methylpentane and dashed lines are theoretical predictions with N B = 1.2 for 2-methylpentane. Symbols are experimental data 3 at 283.15K (squares), 293.15K (triangles), 303.15K (diamomds) and 313.15K (circles). 100

Effect of position where branches grow

80

P[KPa]

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In the framework of SAFTD-LJ-Branch, different positions where branches grow are not taken into account as long as the numbers and types of articulation points stay the same. As shown earlier, this was not an issue for mixtures of propane with 2-methylpentane and 3methylpentane. However, to get some further insight, this section is devoted to understand the nuances between isomers which only differ in the positions where branches grow. We model the mixture of 2-methylpentane and n-octane as well as the mixture of 3methylpentane and n-octane for vapor liquid equilibrium. The results are shown in Fig. 9 and Fig. 10. With N B = 1, the theory predicts the phase behavior well and shows some deviations from experiments 3,5 for both cases. Following the pattern of deviation, we find that a better agreement between prediction and experiments can be obtained when N B is set to around 1.2 for 2-methylpentane and 0.8 for 3methylpentane. This study is interesting as well as intuitive. 2-methylpentane has a ternary carbon atom shielded by two methyl groups while

60

40

20

0 0

0.2

0.4

0.6

0.8

1

x3-methylpentane,y 3-methylpentane

Figure 10: VLE of the mixture 3-methylpentane + noctane at various tmeperatures. Solid lines are theoretical predictions with N B = 1 for 3-methylpentane and dashed lines are theoretical predictions with N B = 0.8 for 3-methylpentane. Symbols are experimental data 3,5 at 283.15K (squares), 293.15K (triangles), 303.15K (diamomds), 313.15K (circles) and 333.15K (crosses).

the ternary carbon atom of 3-methylpentane is surrounded by only one methyl group. A stronger shielding effect for 2-methylpentane leads to a stronger branching effect. Although the most accurate value for N B is dependent on the system studied, we will not try to fit it nor devise a complicated rule of determination for now. SAFTD-LJ-Branch is developed to provide reasonable solutions to ther-


11


A

modynamics of branched molecules even when no experimental data exists for isomers. We see from the above examples that this model can predict the general trend and provide physically meaningful guidance. In this regard, SAFTD-LJ-Branch is a great tool to study the branched molecules. As we see in the case of 2methylpentane and 3-methylpentane, N B = 1 tends to be a good average estimate. In the future, we are interested in a deeper understanding of the relationship between molecular architecture and atoms or groups forming the molecule, potentially leading to better group contribution approaches 56 that explicitly consider molecular architecture.

Page 12 of 17

Expression of A∗s , gs(σ) and gd(σ)

A∗s is proposed by Johnson et al. 48 through correlations of simulation results: P8 ai ρ∗i P6 s As i=1 i + i=1 bi Gi ∗ = (12) As = ∗ Ns kT T where ai , bi are functions of T ∗ , and Gi is function of ρ∗s . The values of gs and gd at contact are also fitted from simulation results. 44,49 gs (σ) = 1 +

5 X 5 X

aij (ρ∗s )i (T ∗ )1−j

(13)

i=1 j=1

4

Conclusion gd (σ) =

This work has a primary focus on thermodynamics of branched molecules. A new equation of state, SAFTD-LJ-Branch is developed through extension of the framework of Marshall and Chapman. 46 The new model is able to predict phase behavior of isomers using a single set of parameters except for the branching parameter, which can be simply obtained from the structure of a molecule. Predictions for pure alkanes and mixtures of alkanes are compared against experimental results with good agreement, especially for the vapor pressures. The accuracy of the predictions is quite surprising considering the simple parameterization scheme and the fact that the branched model adds almost no complexity over the linear model. The theory has strong connections with other popular approaches such as lattice cluster theory and group contribution methods. It also has potential of being applied to associating networks where the effect of branching has been generally neglected.

5

7 X i=1

∗ 1−i

aî (T )

+

5 X 5 X

cij (ρ∗s )i (T ∗ )1−j

i=1 j=1

(14) aij , aî and cij are all sets of constants. Note that in the reference paper 44 aî is basically ai . The hat is put here to distinguish from the previous ai in Eq. 12.

Acknowledgment

We thank the Robert A. Welch Foundation (Grant No. C-1241) for financial support. We thank Kai Langenbach for his insightful comments and suggestions. ACS Paragon Plus Environment

12

Page 13 of 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


References

(10) Dudowicz, J.; Freed, K. F. Effect of monomer structure and compressibility on the properties of multicomponent polymer blends and solutions: 1. Lattice cluster theory of compressible systems. Macromolecules 1991, 24, 5076–5095.

(1) Randic, M. Characterization of molecular branching. J. Am. Chem. Soc. 1975, 97, 6609–6615. (2) Yethiraj, A. Integral equation theory for the surface segregation from blends of linear and star polymers. Comput. Theor. Polym. Sci. 2000, 10, 115–123.

(11) Buta, D.; Freed, K. F.; Szleifer, I. Monte Carlo test of the lattice cluster theory: Thermodynamic properties of binary polymer blends. J. Chem. Phys. 2001, 114, 1424–1431.

(3) Liu, E. K.; Davison, R. R. Vaporliquid equilibriums for the binary systems n-octane with 2-methylpentane, 3methylpentane, and 2, 4dimethylpentane. J. Chem. Eng. Data 1981, 26, 85–88.

(12) Foreman, K. W.; Freed, K. F. Lattice cluster theory of multicomponent polymer systems: Chain semiflexibility and specific interactions. Adv. Chem. Physics, Vol. 103 2007, 335–390.

(4) Chun, S. W.; Rainwater, J. C.; Kay, W. B. Vapor-liquid equilibria of mixtures of propane and isomeric hexanes. J. Chem. Eng. Data 1993, 38, 494–501.

(13) Zeiner, T.; Browarzik, D.; Enders, S. Calculation of the liquid–liquid equilibrium of aqueous solutions of hyperbranched polymers. Fluid Phase Equilib. 2009, 286, 127–133.

(5) Berro, C.; La\ichoubi, F.; Rauzy, E. Isothermal (vapour+ liquid) equilibria and excess volumes of (3-methylpentane+ heptane), of (3-methylpentane+ octane), and of (toluene+ octane). J. Chem. Thermodyn. 1994, 26, 863–869.

(14) Browarzik, D.; Langenbach, K.; Enders, S.; Browarzik, C. Modeling of the branching influence on liquid–liquid equilibrium of binary and ternary polymer solutions by lattice–cluster theory. J. Chem. Thermodyn. 2013, 62, 56–63.

(6) De Loos, T. W.; Poot, W.; Lichtenthaler, R. N. The influence of branching on high-pressure vapor-liquid equilibria in systems of ethylene and polyethylene. J. Supercrit. Fluids 1995, 8, 282–286.

(15) Langenbach, K.; Enders, S. Development of an EOS based on lattice cluster theory for pure components. Fluid Phase Equilib. 2012, 331, 58–79.

(7) Freed, K. F. New lattice model for interacting, avoiding polymers with controlled length distribution. J. Phys. A. Math. Gen. 1985, 18, 871.

(16) Langenbach, K.; Enders, S.; Browarzik, C.; Browarzik, D. Calculation of the high pressure phase equilibrium in hyperbranched polymer systems with the lattice-cluster theory. J. Chem. Thermodyn. 2013, 59, 107–113.

(8) Freed, K. F.; Bawendi, M. G. Lattice theories of polymeric fluids. J. Phys. Chem. 1989, 93, 2194–2203. (9) Dudowicz, J.; Freed, K. F.; Madden, W. G. Role of molecular structure on teh thermodynamic properties of melts, blends, and concentrated polymer solutions. Comparison of Monte Carlo simulations with the cluster theory for the lattice model. Macromolecules 1990, 23, 4803– 4819.

(17) Fischlschweiger, M.; Enders, S. Solid– liquid equilibria of crystalline and semicrystalline monodisperse polymers, taking into account the molecular architecture by application of the lattice cluster theory. Mol. Phys. 2014, 112, 3109–3119.


13


(18) Langenbach, K.; Browarzik, D.; Sailer, J.; Enders, S. New formulation of the lattice cluster theory equation of state for multicomponent systems. Fluid Phase Equilib. 2014, 362, 196–212.

Page 14 of 17

(27) Wertheim, M. S. Fluids with highly directional attractive forces. II. Thermodynamic perturbation theory and integral equations. J. Stat. Phys. 1984, 35, 35–47. (28) Wertheim, M. S. Fluids with highly directional attractive forces. III. Multiple attraction sites. J. Stat. Phys. 1986, 42, 459–476.

(19) Langenbach, K.; Fischlschweiger, M.; Enders, S. Prediction of the solid–liquid– liquid equilibria of linear and branched semi-crystalline poly-ethylene in solutions of diphenyl ether by Lattice Cluster Theory. Mol. Phys. 2016, 114, 2717–2723.

(29) Wertheim, M. S. Fluids with highly directional attractive forces. IV. Equilibrium polymerization. J. Stat. Phys. 1986, 42, 477–492.

(20) Zimmermann, P.; Goetsch, T.; Zeiner, T.; Enders, S. Modelling of adsorption isotherms of isomers using density functional theory. Mol. Phys. 2017, 1–19.

(30) Chapman, W. G. Prediction of the thermodynamic properties of associating Lennard-Jones fluids: Theory and simulation. J. Chem. Phys. 1990, 93, 4299–4304.

(21) Chapman, W. G.; Jackson, G.; Gubbins, K. E. Phase equilibria of associating fluids: chain molecules with multiple bonding sites. Mol. Phys. 1988, 65, 1057– 1079.

(31) Banaszak, M.; Chiew, Y. C.; Radosz, M. Thermodynamic perturbation theory: Sticky chains and square-well chains. Phys. Rev. E 1993, 48, 3760.

(22) Chapman, W. G.; Gubbins, K. E.; Jackson, G.; Radosz, M. SAFT: equation-ofstate solution model for associating fluids. Fluid Phase Equilib. 1989, 52, 31–38.

(32) Gross, J.; Sadowski, G. Perturbed-chain SAFT: An equation of state based on a perturbation theory for chain molecules. Ind. Eng. Chem. Res. 2001, 40, 1244– 1260.

(23) Chapman, W. G.; Gubbins, K. E.; Jackson, G.; Radosz, M. New reference equation of state for associating liquids. Ind. Eng. Chem. Res. 1990, 29, 1709–1721.

(33) Gross, J.; Sadowski, G. Application of the perturbed-chain SAFT equation of state to associating systems. Ind. Eng. Chem. Res. 2002, 41, 5510–5515.

(24) Huang, S. H.; Radosz, M. Equation of state for small, large, polydisperse, and associating molecules. Ind. Eng. Chem. Res. 1990, 29, 2284–2294.

(34) Felipe, J. Thermodynamic behaviour of homonuclear and heteronuclear LennardJones chains with association sites from simulation and theory. Mol. Phys. 1997, 92, 135–150.

(25) Huang, S. H.; Radosz, M. Equation of state for small, large, polydisperse, and associating molecules: extension to fluid mixtures. Ind. Eng. Chem. Res. 1991, 30, 1994–2005.

(35) Gil-Villegas, A.; Galindo, A.; Whitehead, P. J.; Mills, S. J.; Jackson, G.; Burgess, A. N. Statistical associating fluid theory for chain molecules with attractive potentials of variable range. J. Chem. Phys. 1997, 106, 4168–4186.

(26) Wertheim, M. S. Fluids with highly directional attractive forces. I. Statistical thermodynamics. J. Stat. Phys. 1984, 35, 19– 34.

(36) Wertheim, M. S. Thermodynamic perturbation theory of polymerization. J. Chem. Phys. 1987, 87, 7323–7331.


14

Page 15 of 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(37) Phan, S.; Kierlik, E.; Rosinberg, M. L.; Yu, H.; Stell, G. Equations of state for hard chain molecules. J. Chem. Phys. 1993, 99, 5326–5335.

(47) Rowlinson, J. S.; Swinton, F. Liquids and liquid mixtures: Butterworths monographs in chemistry; Butterworth-Heinemann, 2013.

(38) Marshall, B. D.; Chapman, W. G. Higher order classical density functional theory for branched chains and rings. J. Phys. Chem. B 2011, 115, 15036–15047.

(48) Johnson, J. K.; Zollweg, J. A.; Gubbins, K. E. The Lennard-Jones equation of state revisited. Mol. Phys. 1993, 78, 591– 618.

(39) Ghonasgi, D.; Chapman, W. G. A new equation of state for hard chain molecules. J. Chem. Phys. 1994, 100, 6633–6639.

(49) Johnson, J. K.; Mueller, E. A.; Gubbins, K. E. Equation of state for LennardJones chains. J. Phys. Chem. 1994, 98, 6413–6419.

(40) Shukla, K. P.; Chapman, W. G. A twofluid theory for chain fluid mixtures from thermodynamic perturbation theory. Mol. Phys. 1998, 93, 287–293.

(50) Ghonasgi, D.; man, W. G. diatomic and molecules. J. 5662–5667.

(41) Shukla, K. P.; Chapman, W. G. TPT2 and SAFTD equations of state for mixtures of hard chain copolymers. Mol. Phys. 2000, 98, 2045–2052.

Llano-Restrepo, M.; ChapHenry’s law constant for polyatomic Lennard-Jones Chem. Phys. 1993, 98,

(51) Ghonasgi, D.; Chapman, W. G. Prediction of the properties of model polymer solutions and blends. AIChE J. 1994, 40, 878–887.

(42) Dominik, A.; Jain, S.; Chapman, W. G. New Equation of State for Polymer Solutions Based on the Statistical Associating Fluid Theory (SAFT)- Dimer Equation for Hard-Chain Molecules. Ind. Eng. Chem. Res. 2007, 46, 5766–5774.

(52) Linstrom, P. J.; Mallard, W. G. NIST Chemistry webbook; NIST standard reference database No. 69 ; 2001. (53) Pedrosa, N.; Vega, L. F.; Coutinho, J. A. P.; Marrucho, I. M. Phase equilibria calculations of polyethylene solutions from SAFT-type equations of state. Macromolecules 2006, 39, 4240–4246.

(43) Chang, J.; Sandler, S. I. An equation of state for the hard-sphere chain fluid: theory and Monte Carlo simulation. Chem. Eng. Sci. 1994, 49, 2777–2791.

(54) Bymaster, A.; Emborsky, C.; Dominik, A.; Chapman, W. G. Renormalization-group corrections to a perturbed-chain statistical associating fluid theory for pure fluids near to and far from the critical region. Ind. Eng. Chem. Res. 2008, 47, 6264– 6274.

(44) Johnson, J. K. Perturbation theory and computer simulations for linear and ring model polymers. J. Chem. Phys. 1996, 104, 1729–1742. (45) Blas, F. J.; Vega, L. F. Improved vapor– liquid equilibria predictions for LennardJones chains from the statistical associating fluid dimer theory: Comparison with Monte Carlo simulations. J. Chem. Phys. 2001, 115, 4355–4358.

(55) Kay, W. B. The vapor pressures and saturated liquid and vapor densities of the isomeric hexanes. J. Am. Chem. Soc. 1946, 68, 1336–1339.

(46) Marshall, B. D.; Chapman, W. G. Three new branched chain equations of state based on Wertheim’s perturbation theory. J. Chem. Phys. 2013, 138, 174109.

(56) Fredenslund, A. Vapor-liquid equilibria using UNIFAC: a group-contribution method ; Elsevier, 2012.


15


Page 16 of 17

contributes 1 to NB

contributes 3 to NB

NB=0

NB=1

NB=1

NB=2


16

NB=3

Page 17 of 17 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


TOC Graphic 338x190mm (300 x 300 DPI)


Modeling Thermodynamic Properties of Isomeric Alkanes with a New

Recommend Documents